I ´ve recently submitted a paper dealing with Geometric Morphometrics in plants. I used PCA in order to show variation as a previous step to proceed with statistical tests (Procrustes ANOVA, DAs, etc). One reviewer asked us:

"Are there any reasons why PCA analysis limitations (use of variance, linear projection) should not considered important in this analysis ? This kind of question should be addressed, because they are at the heart of any interesting question anyone would ask in performing such analysis : the relations between morphogenesis and variance. It seems unclear to me, after having read the article, why kernel PCA should not be privileged versus PCA"

As I have never read about kernel PCA in GM papers, I want to know if it should be mandatory to compare it with regular PCA in these kind of papers. If not, what do you think the best response should be?

Thanks you all!